Mercurial > matrix-functions

comparison toolbox/dorr.m @ 2:c124219d7bfa draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Re-add the 1995 toolbox after noticing the statement in the ~higham/mctoolbox/ webpage.
author Antonio Pino Robles <data.script93@gmail.com>
date Thu, 07 May 2015 18:36:24 +0200
parents 8f23314345f4
children
comparison
equal deleted inserted replaced
1:e471a92d17be 2:c124219d7bfa
1 function [c, d, e] = dorr(n, theta)
2 %DORR Dorr matrix - diagonally dominant, ill conditioned, tridiagonal.
3 % [C, D, E] = DORR(N, THETA) returns the vectors defining a row diagonally
4 % dominant, tridiagonal M-matrix that is ill conditioned for small
5 % values of the parameter THETA >= 0.
6 % If only one output parameter is supplied then
7 % C = FULL(TRIDIAG(C,D,E)), i.e., the matrix iself is returned.
8 % The columns of INV(C) vary greatly in norm. THETA defaults to 0.01.
9 % The amount of diagonal dominance is given by (ignoring rounding errors):
10 % COMP(C)*ONES(N,1) = THETA*(N+1)^2 * [1 0 0 ... 0 1]'.
11
12 % Reference:
13 % F.W. Dorr, An example of ill-conditioning in the numerical
14 % solution of singular perturbation problems, Math. Comp., 25 (1971),
15 % pp. 271-283.
16
17 if nargin < 2, theta = 0.01; end
18
19 c = zeros(n,1); e = c; d = c;
20 % All length n for convenience. Make c, e of length n-1 later.
21
22 h = 1/(n+1);
23 m = floor( (n+1)/2 );
24 term = theta/h^2;
25
26 i = (1:m)';
27 c(i) = -term*ones(m,1);
28 e(i) = c(i) - (0.5-i*h)/h;
29 d(i) = -(c(i) + e(i));
30
31 i = (m+1:n)';
32 e(i) = -term*ones(n-m,1);
33 c(i) = e(i) + (0.5-i*h)/h;
34 d(i) = -(c(i) + e(i));
35
36 c = c(2:n);
37 e = e(1:n-1);
38
39 if nargout <= 1
40 c = tridiag(c, d, e);
41 end